Unravelling the snowmelt-albedo feedback in Antarctica

Abstract

Along three-quarters of its edge, the Antarctic ice sheet is fringed by ice shelves. These floating extensions of the ice sheet buttress the grounded ice sheet, slowing down the rate at which it is flowing into the ocean. In the recent past, ice shelves have been subject to significant change: in 1995 and 2002, two ice shelves in the Antarctic Peninsula completely disintegrated. As ice shelves float on the ocean, their disappearance does not directly contribute to sea-level rise. However, when ice shelves disappear, so does their buttressing effect on the grounded ice sheet, which causes the ice sheet to flow into the ocean at a faster rate, raising global mean sea level.

An ice shelf is susceptible to a process called 'hydrofracturing'. This occurs when snow melts at the ice-shelf surface, forming melt ponds. The meltwater can fill and deepen pre-existing crevasses and eventually, the crevasse can reach the bottom of the ice shelf. If snowmelt is widespread on the ice-shelf surface, this can potentially destabilise the ice shelf, causing it to disintegrate. Whether snowmelt occurs is determined by the surface energy balance: the sum of all energy fluxes towards and away from the surface. An important energy source is the absorption of solar radiation. The amount of sunlight that is absorbed by the surface is determined by its reflectivity: the surface albedo. Fresh snow has a high albedo (85 %), but when snow melts it becomes darker, reflecting perhaps 70 %, and thus absorbs two times more solar radiation. If the surface absorbs more energy, more snow can melt, darkening the surface even further. This process is known as the snowmelt-albedo feedback.

In this thesis, we explicitly model this process and quantify how strongly it influences snowmelt events on the vast Antarctic ice sheet. In Chapter 3, this is first done for a single location, using meteorological observations from a German research station situated on an ice shelf in coastal Dronning Maud Land, East Antarctica. We find that the snowmelt-albedo feedback enhances surface melt by a factor of 2.5. Data from several automatic weather stations are used in Chapter 4 to evaluate the regional climate model RACMO2. We find that RACMO2 provides reasonable estimates of snowmelt rates: although on average it slightly underestimates snowmelt, the spatial and temporal variability agree well with weather stations and satellite measurements. Finally, in Chapter 5, we use RACMO2 to study the snowmelt-albedo feedback over the entire Antarctic ice sheet. We show that the strength of the snowmelt-albedo feedback is dependent on the frequency of snowfall and the average air temperature.

These results show that it is worth the extra computational costs to explicitly calculate the surface albedo and snowmelt. A simpler snowmelt parameterisation can yield similar Antarctica-wide totals, but lacks the spatial and temporal variability that is related to the snowmelt-albedo feedback. Using an explicit albedo model is therefore crucial if we want to study the future of the Antarctic ice shelves.